Simultaneous single- and multi-contrast super-resolution for brain MRI images based on a convolutional neural network

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Image Processing

Background:

• Magnetic Resonance Imaging (MRI) often yields low-resolution images due to practical limitations like scan time and patient comfort.
• Super-resolution techniques aim to enhance MRI image resolution, categorized into single-contrast (no reference) and multi-contrast (using a reference modality).

Purpose of the Study:

• To introduce a novel deep convolutional neural network (CNN) capable of performing both single- and multi-contrast super-resolution for MRI.
• To simultaneously reconstruct high-resolution images from low-resolution inputs, addressing limitations of existing methods.

Main Methods:

• Development of a deep convolutional neural network architecture designed for simultaneous single- and multi-contrast MRI super-resolution.
• Utilizing both synthetic and real brain MRI datasets for model training and validation.

Main Results:

• The proposed CNN model demonstrated superior performance compared to current state-of-the-art MRI super-resolution techniques.
• Quantitative improvements were observed in peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).
• Enhanced visual quality of reconstructed high-resolution MRI images was achieved.

Conclusions:

• The developed deep convolutional neural network effectively performs simultaneous single- and multi-contrast MRI super-resolution.
• The model offers significant improvements in both objective quality metrics and subjective visual assessment for brain MRI.
• This approach holds promise for advancing diagnostic capabilities through enhanced MRI resolution.